Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 1 Impacts of absorbing aerosol deposition on snowpack and hydrologic 2 cycle in the Rocky Mountain region based on variable-resolution 3 CESM (VR-CESM) simulations 4 Chenglai Wu1,2, Xiaohong Liu1,*, Zhaohui Lin2,3, Stefan R. Rahimi-Esfarjani1, and 5 Zheng Lu1 6 1Department of Atmospheric Science, University of Wyoming, Laramie, Wyoming, 7 USA 8 2International Center for Climate and Environment Sciences, Institute of Atmospheric 9 Physics, Chinese Academy of Sciences, Beijing, China 10 3University of Chinese Academy of Sciences, Beijing, China 11 12 *Corresponding to: 13 Xiaohong Liu 14 Department of Atmospheric Science 15 University of Wyoming 16 Dept. 3038, 1000 East University Avenue 17 Laramie, WY 82071 18 Email: [email protected]. 19 20 1 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 21 Abstract 22 Deposition of light-absorbing aerosols (LAAs) such as black carbon (BC) and dust 23 onto snow cover has been suggested to reduce the snow albedo, and modulate the 24 snowpack and consequent hydrologic cycle. In this study we use the 25 variable-resolution Community Earth System Model (VR-CESM) with a regionally 26 refined high-resolution (0.125º) grid to quantify the impacts of LAAs in snow in the 27 Rocky Mountain region during the period of 1981-2005. We first evaluate the model 28 simulation of LAA concentrations both near the surface and in snow, and then 29 investigate the snowpack and runoff changes induced by LAAs in snow. The model 30 simulates similar magnitudes of near-surface atmospheric dust concentrations as 31 observations. Although the model underestimates near-surface atmospheric BC 32 concentrations, the simulated BC-in-snow concentrations are overall comparable to 33 observations. Regional mean surface radiative effect (SRE) due to LAAs in snow 34 reaches up to 0.6-1.7 W m-2 in spring, and dust contributes to about 21-43% of total 35 SRE. Due to positive snow-albedo feedbacks induced by the LAAs’ SRE, snow water 36 equivalent reduces by 2-50 mm and snow cover fraction by 5-20% in the two regions 37 around the mountains (Eastern Snake River Plain and Southwestern Wyoming), 38 corresponding to an increase of surface air temperature by 0.9-1.1°C. During the snow 39 melting period, LAAs accelerate the hydrologic cycle with monthly runoff increases 40 of 0.15-1.00 mm day-1 in April-May and reductions of 0.04-0.18 mm day-1 in 41 June-July in the mountainous regions. Of all the mountainous regions, Southern 2 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 42 Rockies experience the largest reduction of total runoff by 15% during the later stage 43 of snow melt (i.e., June and July). Our results highlight the potentially important role 44 of LAA interactions with snowpack and subsequent impacts on the hydrologic cycles 45 across the Rocky Mountains. 46 47 1. Introduction 48 Water resources are essential to human society and economic development as 49 well as ecosystems in the western United States. Most of primary water resources in 50 the inland western U.S. come from the Rocky Mountains’ snowpack (Serreze et al., 51 1999). Therefore, to develop the water resource management strategy, it is necessary 52 to know the information of snow accumulation and snowmelt timing. Climate change 53 is an important factor influencing the snowpack in the Rocky Mountain region, as has 54 been shown in many previous studies (e.g., Abatzoglou, 2011; Pederson et al., 2011; 55 Rhoades et al., 2017). Another important factor is the light-absorbing aerosols (LAAs, 56 e.g., black carbon (BC), organic carbon (OC), and dust) in snow (e.g., Flanner et al., 57 2007; Painter et al., 2007; Qian et al., 2015; Yasunari et al., 2015). Previous studies 58 have shown that LAAs in snow can significantly reduce the surface albedo (often 59 known as snow darkening effect (SDE)), modify the surface energy budget and 60 snowmelt, and lead to the modification of hydrologic cycles (e.g., Warren and 61 Wiscombe, 1980; Hansen and Nazarenko, 2004; Flanner et al., 2007, 2009; Painter et 62 al., 2007, 2010; Qian et al., 2009, 2011; Yasunari et al., 2015). Moreover, the 3 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 63 LAAs-induced snow albedo reduction may initiate positive feedback processes, which 64 can amplify the reduction of snowpack (e.g., Flanner et al., 2009; Qian et al., 2009). 65 In past decades modeling studies have been undertaken to quantify the impacts 66 of SDE by LAAs (e.g., Flanner et al., 2007; Qian et al., 2009; Oaida et al., 2015; 67 Yasunari et al., 2015). Generally the models they developed have the ability to 68 simulate the temporal evolution of snow albedo under the influence of LAAs in snow. 69 These studies have enhanced our understanding of the spatial and temporal variations 70 of climate forcings and responses due to LAAs in snow from regional scales (e.g., 71 Qian et al., 2009; Oaida et al., 2015) to global scales (e.g., Flanner et al., 2007; 72 Yasunari et al., 2015). For example, the impacts of LAAs in snow are stronger in 73 regions with considerable snow cover and sufficient LAAs deposition (e.g., Arctic, 74 Northeast China, Tibetan Plateau, and western U.S.) than in other regions, and they 75 are largest during the snowmelt period due to the positive snow-albedo feedback. 76 However, as also mentioned in these studies, reliable quantification of impacts of 77 LAAs in snow is hindered by the model deficiencies in simulating the snowpack and 78 aerosol cycles, with additional uncertainties induced by the parameterization of 79 snow-aerosol-radiation interactions. 80 In particular, previous studies used the coarse-resolution global climate 81 models (GCMs) or high-resolution regional climate models (RCMs) to quantify the 82 impacts of LAAs in snow. However, there are weaknesses in either coarse-resolution 83 GCMs or RCMs. Both snowfall and snow accumulation depend on the temperature 4 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 84 and precipitation, and thus distribution of snowpack depends strongly on topographic 85 variablility. Current GCMs with a typical horizontal resolution of 1° to 2° cannot 86 resolve the snowpack over the regions with complex terrains (e.g., Rocky Mountains) 87 due to the coarse resolution (Rhoades et al., 2016; Wu et al., 2017), which impedes 88 the reliable quantifications of SDE by LAAs in mountainous regions (e.g., Flanner et 89 al., 2007; Yasunari et al., 2015). For RCMs, they can simulate the snowpack more 90 accurately but are not able to simulate the global transport of aerosols to the focused 91 region except that aerosol transport along the boundary is prescribed (e.g., Qian et al., 92 2009; Oaida et al., 2015). Moreover, LAAs in snow may also influence the climate 93 beyond the focused region (e.g., Yasunari et al., 2015), which cannot be accounted for 94 in RCMs. Variable resolution GCMs (VR-GCMs) can overcome these weaknesses of 95 either coarse-resolution GCMs or RCMs and serve as a better tool to quantify the 96 impacts of LAAs in snow. Although GCMs with globally uniform high resolutions 97 (10-30 km) may be an ideal tool to simulate the snowpack and snow-aerosol-radiation 98 interactions, they are not widely applied due to the constraints of computational 99 resources (e.g., Haarsma et al., 2016). Instead, using VR-GCMs is a more economic 100 approach and has gained increasing utility in recent years (e.g., Zarzycki et al, 2014a, 101 b; Sakaguchi et al., 2015). 102 A variable-resolution version of the Community Earth System Model 103 (VR-CESM) has been developed (Zarzycki et al., 2014a, b). With a refined high 104 resolution, VR-CESM has shown significant improvements of the Atlantic tropical 5 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 105 storms (Zarzycki and Jablonowski, 2014) and South America orographic precipitation 106 (Zarzycki et al., 2015). The model has also been used in the regional climate 107 simulations in western U.S., and results show that VR-CESM is capable of 108 reproducing the spatial patterns and the seasonal evolution of temperature, 109 precipitation, and snowpack in Sierra Nevada (Huang et al., 2016; Rhoades et al., 110 2016) and in Rocky Mountains (Wu et al., 2017). In particular, VR-CESM simulates 111 reasonably the magnitude of snow water equivalent, the timing of snow water 112 equivalent peaks, and the duration of snow cover in the Rocky Mountains by 113 comparing against the Snow Telemetry (SNOTEL) and MODIS (Moderate 114 Resolution Imaging Spectroradiometer) snow cover observations (Wu et al., 2017). 115 Following the evaluation study of Wu et al. (2017), here we use VR-CESM to 116 investigate the impacts of LAAs in snow (BC and dust) on the snowpack and 117 hydrologic cycles over the Rocky Mountains. By comparing the two VR-CESM 118 simulations with and without considering the impacts of LAAs in snow, we examine 119 the changes in surface radiative transfer, temperature, snowpack, and runoff induced 120 by LAAs in snow. To our knowledge, it is the first time that VR-CESM is applied for 121 the study of LAAs in snow. Our results will demonstrate that VR-CESM is skillful for 122 this kind of research. 123 The remainder of the paper is organized as follows. Section 2 introduces the 124 model and experimental design. Section 3 describes the observation data used for 125 validation of model simulations of aerosol fields in the surface air and in snow. 6 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 126 Section 4 presents the evaluation of aerosols fields, followed by their surface radiative 127 effect (SRE), as well as the change of surface temperature, snowpack, and runoff 128 induced by LAAs in snow. Discussion and conclusions are given in section 5. 129 2. Model and experimental design 130 The model used in this study is VR-CESM, a version of CESM (version 1.2.0) 131 with the variable-resolution capability (Zarzycki et al., 2014a, b). CESM is a 132 state-of-the-art Earth system modeling framework that allows for for investigations of 133 a diverse set of Earth system interactions across multiple time and space scales 134 (Hurrell et al., 2013). CESM uses the Community Atmosphere Model version 5 135 (CAM5) for the atmospheric component (Neale et al., 2010). The variable-resolution 136 capability is implemented into the Spectral Element (SE) dynamic core of CAM5. 137 The SE dynamic core uses a continuous Galerkin spectral finite-element method 138 designed for fully unstructured quadrilateral meshes, and has demonstrated 139 near-optimal (close to linear) parallel scalability on tens of thousands of cores (Dennis 140 et al., 2012). This enables the model to run efficiently on decadal to multi-decadal 141 time scales. For the land component, CESM uses the Community Land Model version 142 4 (CLM4). CLM4 can be run at the same horizontal resolutions as CAM5 and thus 143 also benefit from the variable-resolution capability of CAM5. 144 CESM also includes advanced physics for CAM5 (Neale et al., 2010) and 145 CLM4 (Oleson et al., 2010). The CAM5 physics suite consists of shallow convection 146 (Park and Bretherton, 2009), deep convection (Zhang and McFarlane, 1995; Richter 7 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 147 and Rasch, 2008), cloud microphysics (Morrison and Gettelman 2008) and 148 macrophysics (Park et al. 2014), radiation (Iacono et al. 2008), and aerosols (Liu et al., 149 2012). For aerosols, a modal aerosol module (MAM) is adopted to represent the 150 internal and external mixing of aerosol components such as BC, OC, sulfate, 151 ammonium, sea salt, and mineral dust (Liu et al., 2012). CLM4 physics includes a 152 suite of parameterizations for land-atmosphere exchange of water, energy and 153 chemical compounds. In particular, CLM4 explicitly represents the snowpack (snow 154 accumulation and melt) by a snow model and its coupling with the SNow, Ice and 155 Aerosol Radiation (SNICAR) model for snow-aerosol-climate interactions (Flanner et 156 al., 2007). SNICAR incorporates a two-stream radiative transfer solution of Toon et 157 al. (1989) to calculate the snow albedo and the vertical absorption profile from solar 158 zenith angle, albedo of the substrate underlying snow, mass concentrations of 159 atmospheric-deposited aerosols (BC and dust), and ice effective grain size (r). r is e e 160 simulated with a snow aging routine (Oleson et al., 2010). SNICAR is compatible 161 with the new modal aerosol module of CAM5 (Liu et al., 2012) in the treatment of 162 aerosol deposition (Flanner et al., 2012). As our knowledge of OC optical properties 163 is limited, the impact of absorbing OC on snow albedo is not included in the standard 164 CLM4 and thus not considered in this study. 165 For the high-resolution modeling, we have designed a variable-resolution grid 166 that transits from global quasi-uniform 1º resolution to a refined 0.125º resolution in 167 the Rocky Mountains (Figure 1a). The variable-resolution grid is the same as that 8 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 168 used in Wu et al. (2017), and is generated by the open-source software package called 169 SQuadGen (Ullrich, 2014). A topographical dataset for this variable-resolution grid is 170 also generated accordingly by the National Center for Atmospheric Research (NCAR) 171 global model topography generation software called NCAR_Topo (v1.0) (Lauritzen et 172 al., 2015) as described in Wu et al. (2017). As shown in Figure 1b, the high-resolution 173 grids resolve well the variations of terrain in the Rocky Mountains. Note that the 174 standard CESM using the coarse 1º grids cannot resolve the fine-scale variations of 175 terrain in the Rocky Mountains (see Figure 2 of Wu et al. (2017)). In Wu et al. (2017), 176 we have shown that VR-CESM performs well in the simulation of regional climate 177 patterns, including spatial distributions and seasonal evolution of temperature, 178 precipitation, and snowpack in the Rocky Mountain region. In this study, we further 179 apply VR-CESM to simulate the SDE of LAAs and its impacts on snowpack and 180 hydrologic cycles in the Rocky Mountains. 181 VR-CESM is run in the coupled land-atmosphere mode with prescribed 182 observed monthly 1º×1º sea surface temperature and sea ice coverage (Hurrell et al., 183 2008), following the Atmospheric Model Intercomparison Project (AMIP) protocols 184 (Gates, 1992). The simulation period is from 1979 to 2005, and the results for the last 185 25 years (1981-2005) are used for the analysis shown below. Historical greenhouse 186 gas concentrations, and anthropogenic aerosol and precursor gas emissions are 187 prescribed from the datasets of Lamarque et al. (2010). In particular, the BC 188 emissions consist of various sources, including domestic, energy, transportation, 9 Atmos.Chem.Phys.Discuss.,https://doi.org/10.5194/acp-2017-799 ManuscriptunderreviewforjournalAtmos.Chem.Phys. Discussionstarted:4September2017 (cid:13)c Author(s)2017.CCBY4.0License. 189 waste, shipping, and wildfire (forest and grass fires) emissions. We note that the 190 horizontal resolution for BC emission used in this study is 1.9×2.5º. The relatively 191 coarse resolution of BC emission may partly explain the model’s bias in the 192 simulation of BC concentrations near the surface and in snow across regions where 193 local BC sources can contribute significantly to the observed BC concentrations, as 194 will be discussed in section 4. 195 For dust aerosol, the emission flux is calculated interactively in the model at 196 each time step by a dust emission scheme (Oleson et al., 2010). The dust emission 197 flux is calculated from the friction velocity, threshold friction velocity, atmospheric 198 density, clay content in the soil, areal fraction of exposed bare soil, and source 199 erodibility (Oleson et al., 2010; Wu et al., 2016). Due to the large uncertainty in 200 modeled dust emission, the dust emission scheme also adopts a posterior tuning factor 201 (T) to simulate the reasonable dust emission amount. With the increase of model 202 resolution, VR-CESM produces much higher dust concentrations compared to the 203 observations (section 3) in North America if T used in the standard CESM with 204 quasi-uniform 1º resolution is used. Therefore, for VR-CESM, T is reduced by a 205 factor of 2.6 to produce the similar magnitudes of near-surface dust concentrations as 206 the observations, as will be shown in section 4.1. Note that such a reduction of T is 207 only applied in North America, since other continents have a resolution of 208 quasi-uniform 1º, the same as in the standard CESM. 10
Description: